The April 25, 1992, Cape Mendocino earthquake (MS 7.1) has renewed speculation about the nature of subduction along the Cascadia subduction zone and the associated seismic hazard. This event may represent the first large (M>6) thrust event along the entire Cascadia subduction zone in historic times (last 200 years). We analyze long-period surface waves and broadband body waves in order to estimate the mainshock source parameters. We also examine broadband body waves from the nearby 1991 Honeydew earthquake (M 6) in order to assess the contributions of both rupture complexity and unmodeled source and receiver structure in the Cape Mendocino waveforms. From both body and surface wave inversions, as well as forward modeling of body waves, we estimate a best double couple mechanism for the Cape Mendocino earthquake (strike=330¿10¿, dip=12¿2¿, rake=75¿15¿, seismic moment=1.93¿1019 N m, and Mw=6.8). This mechanism contains a significant component of slip in the estimated direction of Gorda-North America plate convergence. Although this earthquake had suitable geometry for relieving strain accumulated by Gorda-North America plate convergence, we cannot resolve whether it occurred on the interplate megathrust or on a fault within the overriding accretionary prism. We find evidence for southwest (offshore) rupture toward an azimuth of 240¿. In addition, we find evidence for early aftershocks in both the teleseismically recorded body waves and in locally recorded strong motions. We model one aftershock delayed ~13 s from the mainshock with a mechanism that is different from the mainshock mechanism but is consistent with the north-south trending, horizontal compression found offshore within the Gorda plate. We postulate that this aftershock and two additional large, strike-slip aftershocks that ruptured the Gorda plate within 24 hours of the mainshock were caused by the transfer of stress accumulated across the Cascadia subduction zone and accretionary prism far offshore, to the Gorda plate, where it reduced the normal stress across NW-SE oriented faults, triggering failure. The complexity of fault interactions near the Mendocino triple junction needs to be understood before potential seismic hazards of the southern Cascadia subduction zone can be quantified. ¿ American Geophysical Union 1996